Even if you think the Blockchain itself is valuable, the "proof of work" is almost entirely wasted in the end. Some trillions of hash computations are thrown away for each one that is used.
Even as a blockchain non-fan, those computations are necessary to produce the desirable thing, and the same could be said for protein folding. There are a lot of great and devastating criticisms of blockchains, but the environmental concern is more of a proxy or even dogwhistle for a general political objection to cryptocurrencies - even reasonable objections in the context of their worldview. Greenwashing those objections seems like a canard.
I think you two are disagreeing on what "the desirable thing" means. You're saying it's creating the bitcoin/blockchain which is the desirable result. But they're saying that "desirable result" here is actually just computing a SHA-256 below a certain threshold. Sure you get the blockchain as a result, but does it really need to be SHA-256? As long as you have a "hash" function with a statistically predictable amount of CPU usage to generate a "successful" result, that function could maybe be something more meaningful. I have no idea if you could quantify protein folding into some sort of statistically predictable successful/unsuccessful CPU process.